])
stim_side = np.array([
np.concatenate(
(stim_sides_arr[k], np.zeros(max_len - len(stim_sides_arr[k]))))
for k in range(len(stim_sides_arr))
])
session_uuids = np.array(session_uuids)
# Fit previous stimulus side model
if 'stimside' in TARGET:
model = exp_stimside(
join(SAVE_PATH, 'Behavior', 'exp_smoothing_model_fits/'),
session_uuids, subject, actions, stimuli, stim_side)
elif 'prevaction' in TARGET:
model = exp_prev_action(
join(SAVE_PATH, 'Behavior', 'exp_smoothing_model_fits/'),
session_uuids, subject, actions, stimuli, stim_side)
model.load_or_train(nb_steps=2000, remove_old=REMOVE_OLD_FIT)
params = model.get_parameters(parameter_type='posterior_mean')
if 'prior' in TARGET:
target = model.compute_signal(signal='prior',
act=actions,
stim=stimuli,
side=stim_side,
parameter_type='posterior_mean',
verbose=False)['prior']
elif 'prederr' in TARGET:
target = model.compute_signal(
signal='prediction_error',
act=actions,
# Load in data
eid = decoding_result.loc[i, 'eid']
probe = decoding_result.loc[i, 'probe']
region = decoding_result.loc[i, 'region']
subject = decoding_result.loc[i, 'subject']
all_eids = np.sort(
decoding_result.loc[decoding_result['subject'] == subject,
'eid'].unique())
print('Processing %d of %d (region %s)' % (i + 1, N_SESSIONS, region))
# Load in fitted behavioral model
if TARGET == 'prior-prevaction':
model = exp_prev_action(
join(SAVE_PATH, 'Behavior', 'exp_smoothing_model_fits/'), all_eids,
subject, np.array([np.array(None)] * all_eids.shape[0]),
np.array([np.array(None)] * all_eids.shape[0]),
np.array([np.array(None)] * all_eids.shape[0]))
elif TARGET == 'prior-stimside':
model = exp_stimside(
join(SAVE_PATH, 'Behavior', 'exp_smoothing_model_fits/'), all_eids,
subject, np.array([np.array(None)] * all_eids.shape[0]),
np.array([np.array(None)] * all_eids.shape[0]),
np.array([np.array(None)] * all_eids.shape[0]))
try:
model.load_or_train(nb_steps=2000, remove_old=False)
except:
continue
params = model.get_parameters(parameter_type='posterior_mean')
# Get priors per trial